AI is being used more and more in these large companies, as well as in startups. According to McKinsey, almost 80% of the world’s largest companies say they’re already using AI in at least one business area. In the UK, over a third of mid-sized firms have also started using these tools.
Many businesses have introduced tools like chatbots or employee support assistants. These are easier to roll out but often bring small changes. The more powerful uses of AI such as full automation of certain tasks are mostly stuck in early trial stages.
Between May 2022 and May 2025, job vacancies went down by 43%, according to McKinsey. Roles that are more exposed to AI, such as software, legal and media jobs, saw a 38% decrease in adverts. Jobs with little AI exposure dropped by 21%. Overall unemployment is expected to go from 4.6% to 4.8% before the end of the year.
Can Companies Do Anything Differently?
Experts often bring up how pulling back on hiring may hurt businesses in the long run. If entry level roles disappear now, it becomes harder to build strong teams in the future. McKinsey says companies should rethink how they use AI as well as how they keep hiring and training people.
Dr Lisa Blue, from Eastern Kentucky University, says AI should be a tool instead of a replacement. Historians, writers, teachers and mathematicians may use AI to support their work, but not lose their jobs. Others agree.
Now, what about startup founders who built their models around high risk jobs? As you know, Microsoft recently released research on jobs at risk and safe from AI, and this makes startups wonder where they stand. Experts share their thoughts and advice…
Our Experts:
- Darren Linden, CEO, Modu
- Sam Littlefield, president & CEO, Littlefield Agency
- Brandon Rollins, Founder & CEO, Pangea Marketing Agency
- Ryan Zhang, Founder and CEO, Notta.ai
- Peter Wood, CTO, Spectrum Search
- Dominic Monkhouse, Business Coach and Founder, Monkhouse & Company
Darren Linden, CEO, Modu
“Here’s our advice: don’t panic. Headlines about GenAI replacing jobs might cite the Microsoft paper but you can bet hardly anyone’s read it.
“What the study does show is that people ask AI for help with writing or gathering information, not that AI completes those tasks well.
“Crucially, it tracks activities, not outcomes. It doesn’t assess whether responses were usable, correct or valuable. And we know GenAI compounds errors over longer conversations, so many of these tasks will have been performed poorly.
“It’s observational, not predictive and they looked at only one tool (Bing Copilot). There’s no evidence of organisational change, displacement or economic impact, only that knowledge workers are experimenting.
“If your product serves these roles, this is a signal of interest, not threat. Double down on product market fit and talk to customers. Opportunity lies in building tools that enhance human work, not replacing it with AI guesswork.”
Gene Genin, CEO and Founding Partner, OEM Source
“Even if your startup is relying on an addressable soon to be flattened (by AI), ask yourself this simple question namely; are you actually solving a real problem or just holding on to a job description? It is possible to distinguish market timing and wishful thinking. We have rejected whole systems in our operation because they were using solvers which were solving constraints that were yesterday and not regarding the constraint today as far as the behavior is concerned.
“I have sat in rooms where a founder is pitching a product which encapsulates an outdated process and hopes that AI does not move as fast as it does or that customers will not realize. It is a lousy bet. Technology is indifferent to what you were intending to construct. it worries what yet remains.
“So cut the sentiment. Begin at the beginning. Go on what remains vital when the role is lost. In case there is nothing left the market has already responded on your behalf.”
Sam Littlefield, president & CEO, Littlefield Agency
“Startups built around roles flagged as “at risk” from AI shouldn’t see this as a death sentence, but rather a call to sharpen their value proposition. Founders have to zoom out and consider the core problem their product solves and how to relay that accordingly to their target audience. The goal should be building for outcomes and workflows, not specific job functions. We talk about it all the time at our agency, we will have so many “thinkers” in the future as opposed to as many “doers”.
“Founders can differentiate by going beyond basic automation. Generic tools will flood the market, but the real opportunity lies in layering in domain expertise, compliance-readiness, and workflow integration. Rather than replacing humans, products should enhance their performance through efficiency, accuracy, or scale. AI won’t always replace jobs outright; it will transform them. Startups that facilitate this transformation are well-positioned.
“It’s also critical to track AI progress and regulation closely. Some roles flagged as “at risk” may persist due to legal or societal constraints, creating niches where thoughtful innovation thrives. Lastly, build in optionality. Keep your roadmap flexible, test new use cases, and have a pivot mindset. Founders who succeed here won’t necessarily be the ones who race to automate, but the ones who thoughtfully adapt to what your customers need next.”
More from Artificial Intelligence
Brandon Rollins, Founder & CEO, Pangea Marketing Agency
“I’m in this boat right now since much of my agency’s work falls into copywriting and copyediting, web development, market research, and similar categories. To me, it seems like the greatest way to manage risk is to upskill in areas that LLMs tend to be weak: fact-checking, reasoning, critical thinking, and so on.
“Basic writing, proofreading, first-draft coding, and low-effort research tasks are being quickly rendered into commodities by AI. But higher-order reasoning is a place where AIs currently struggle and where improvement is likely to be slow.
“To me, the people at a greatest risk in this market are early-career professionals in these fields. For them, I think it would be best to try to focus on upskilling into higher-order reasoning as fast as possible while simultaneously building up skills using AIs. As fast as the tech is, it’s going to take people a long time to adapt to it (as with smartphones, as with PCs), so having fluency early and knowing the strengths and weaknesses of AI as a tool will be tremendously beneficial for years to come.”
Ryan Zhang, Founder and CEO, Notta.ai
“Founders whose ideas relate to jobs most exposed to AI-driven automation need not freak out. Rather, consider it as a strategic step in highlighting where disruption in the market gives way to new opportunities. The question is not so much whether AI will affect these jobs, but how you can develop solutions to augment this change rather than fight it.
“Most of the jobs listed are not disappearing completely but shifting to higher-level work where humans can use their imagination, emotional intelligence, subtle decision-making, and relationship-building. Build on these unique human strengths rather than trying to automate them away.
“The biggest strategic error is competing with AI on AI’s terms. If your business model depends on doing something more cheaply or quickly than AI, you’re building on quicksand. Instead, build tools that allow humans to succeed in their new roles.
“The individuals actually doing these jobs know exactly how their work is transforming and where they require assistance. They will inform you what tasks they’re content to have AI accomplish and what tasks they can do with more effective tools to execute themselves.
“The startups that thrive won’t be the ones fighting AI disruption. They’ll be the ones helping humans adapt to and leverage it.”
Peter Wood, CTO, Spectrum Search
“If your startup idea is built around roles that are now seen as high risk because of AI, that doesn’t mean it’s finished. It’s just a signal to re-evaluate where human input still holds value. I treat risk as a pointer, not something to avoid. If AI can handle most of a task, focus instead on the part it cannot do. That often includes judgement, trust, creativity and subtlety. Those are the areas where new products can stand out.
“Founders should think in terms of workflows rather than job titles. Look closely at the processes your product touches and ask which steps could be automated and which still need a person involved. If your business depends on a role that may soon disappear, that is not a reason to abandon the idea. It is a chance to reshape it with a better understanding of where AI fits and where it falls short. That shift can sometimes reveal a broader, more scalable opportunity than the original concept ever offered.”
Dominic Monkhouse, Business Coach and Founder, Monkhouse & Company
“AI eats McDonald’s before Michelin.”
“If your startup depends on a role that’s just been flagged as ‘at risk’ by AI research, the first question is: are you building a McDonald’s or a Michelin-starred restaurant?
“Generative AI is brilliant at anything standardised, repeatable, and pattern-driven – think McDonald’s: the same burger, a million times over. Startups that rely on formulaic tasks – copywriting mills, basic support services, generic analytics – are first in line for automation.
“But AI can’t do taste. It can’t do real experience, human trust, or the little flashes of creative intuition that make something exceptional. That’s the Michelin side of the spectrum. If your product depends on judgement, storytelling, or high-touch human nuance, AI is more of a sous-chef than a replacement.
My advice to founders:
“Augment, don’t panic. Use AI to strip out the grunt work so your team can focus on the bits machines can’t touch – taste, nuance, and human connection.
“Harden your value proposition. Make sure there’s something unautomatable in what you offer – speed, trust, originality, or an experience people genuinely care about.
“Pivot early if you’re a McDonald’s. If your core idea is formulaic and easily cloned by AI, don’t wait to be disrupted. Move into areas where human + AI collaboration adds 10x value.
“The takeaway: generative AI doesn’t kill jobs so much as it kills mediocrity. If your idea is built on repeatable output, rethink it. If it’s built on craft, experience, and trust, AI is just another tool in your kitchen.”
